I was reading the docstrings for unicode.encode() and
unicode.decode() and they were making no sense to me
since I kept expecting ``unicode("hi",
"utf-8").encode("utf-16")`` to return a unicode object,
just with an internal representation of UTF-16.  Then I
started playing with them some more and I realized the
methods encoded and decoded into and out of a byte
stream.  Then it made sense.

Attached is a patch to mention that they are working
with byte streams.  I also capitalized the first words
in all the sentences.

Logged In: YES 
user_id=55188

That's not true. While the most unicode codecs are doing
such, some doesn't.  It's up to each codec's design what it
gets in its unicode.decode and what it returns in its
unicode.encode.

eg:

>>> u'hello, \uAC00!'.encode('breaker')
u'hello, \xea\xb0\x80!'
>>> _.decode('breaker')
u'hello, \uac00!'


class Codec(codecs.Codec):
....def encode(self, data, errors='strict'):
........return
data.encode('utf-8').decode('iso8859-1'),len(data)
....def decode(self, data, errors='strict'):
........return data.encode('iso8859-1').decode('utf-8'),
len(data)

Logged In: YES 
user_id=357491

That figures.  OK, closed as rejected.